Improved information about the vertical location and extent of monolayer clouds from POLDER3 measurements in the oxygen A-band

[摘要] This paper describes new advances in the exploitation of oxygen A-bandmeasurements from POLDER3 sensor onboard PARASOL, satellite platform withinthe A-Train. These developments result from not only an account of the dependence ofPOLDER oxygen parameters to cloud optical thickness τ and to the scene'sgeometrical conditions but also, and more importantly, from the finerunderstanding of the sensitivity of these parameters to cloud verticalextent. This sensitivity is made possible thanks to the multidirectionalcharacter of POLDER measurements. In the case of monolayer clouds thatrepresent most of cloudy conditions, new oxygen parameters are obtained andcalibrated from POLDER3 data colocalized with the measurements of the twoactive sensors of the A-Train: CALIOP/CALIPSO and CPR/CloudSat. From aparameterization that is (μ_s, τ) dependent, withμ_s the cosine of the solar zenith angle, a cloud top oxygenpressure (CTOP) and a cloud middle oxygen pressure (CMOP) are obtained, whichare estimates of actual cloud top and middle pressures (CTP and CMP).Performances of CTOP and CMOP are presented by class of clouds following theISCCP classification. In 2008, the coefficient of the correlation betweenCMOP and CMP is 0.81 for cirrostratus, 0.79 for stratocumulus, 0.75 for deepconvective clouds. The coefficient of the correlation between CTOP and CTP is0.75, 0.73, and 0.79 for the same cloud types. The score obtained by CTOP,defined as the confidence in the retrieval for a particular range of inferredvalue and for a given error, is higher than the one of MODIS CTP estimate.Scores of CTOP are the highest for bin value of CTP superior in numbers. Forliquid (ice) clouds and an error of 30 hPa (50 hPa), the scoreof CTOP reaches 50% (70%). From the difference between CTOP andCMOP, a first estimate of the cloud vertical extent h is possible. A secondestimate of h comes from the correlation between the angular standarddeviation of POLDER oxygen pressure σ_PO2 and the cloudvertical extent. This correlation is studied in detail in the case of liquidclouds. It is shown to be spatially and temporally robust, except for cloudsabove land during winter months. The analysis of the correlation's dependenceon the scene's characteristics leads to a parameterization providing h fromσ_PO2. For liquid water clouds above ocean in 2008, themean difference between the actual cloud vertical extent and the oneretrieved from σ_PO2 (from the pressure difference)is 5 m (−12 m). The standard deviation of the mean difference isclose to 1000 m for the two methods. POLDER estimates of the cloudgeometrical thickness obtain a global score of 50% confidence for arelative error of 20% (40%) of the estimate for ice (liquid) clouds over ocean. These results need to be validated outside of theCALIPSO/CloudSat track.

[学科分类] 几何与拓扑